How good is the evidence that VBench-2.0 scores track human judgment, which systems does it say lead and by how much, and what judge-bias, saturation or protocol risks do the authors and repo admit or leave?
The human-alignment evidence is a model-level comparison over only four systems per dimension. I recomputed the table's 'Correlation' row as Pearson r over those four points (matches to within rounding), while the figure caption says Spearman; with n=4 rank agreement is far weaker on some dimensions (0.4 for Human Anatomy and Human Interaction, average ranks, no significance test). Several dimensions have tiny score spreads between systems, and the authors admit VLM judges are unreliable for Human Anatomy and Motion Rationality. Between arXiv v1 and v2 the Instance Preservation method and values changed for the same four systems (about 71-76% to 82-94%). Leader statements: per Table II no system leads everywhere; there is no paper aggregate, and leaderboard Total Scores were not retrievable.
Structured fields
{
"citations": [
{
"url": "https://arxiv.org/html/2503.21755v2",
"note": "Table IV (table element 5)"
},
{
"url": "https://github.com/Vchitect/VBench/tree/fd18b3d055cb0fc6f066ca90fe2c3c8cbb698490/VBench-2.0",
"note": "Table IV (table element 5)"
},
{
"url": "https://arxiv.org/html/2503.21755v1",
"note": "Table IV (table element 5)"
},
{
"url": "https://huggingface.co/spaces/Vchitect/VBench_Leaderboard",
"note": "Table IV (table element 5)"
},
{
"url": "https://vchitect.github.io/VBench-2.0-project/",
"note": "Table IV (table element 5)"
}
]
}Evidence claims
- [sourced_fact] The human-alignment evidence is a model-level comparison over only four systems per dimension. I recomputed the table's 'Correlation' row as Pearson r over those four points (matches to within rounding), while the figure caption says Spearman; with n=4 rank agreement is far weaker on some dimensions (0.4 for Human Anatomy and Human Interaction, average ranks, no significance test). Several dimensions have tiny score spreads between systems, and the authors admit VLM judges are unreliable for Human Anatomy and Motion Rationality. Between arXiv v1 and v2 the Instance Preservation method and values changed for the same four systems (about 71-76% to 82-94%). Leader statements: per Table II no system leads everywhere; there is no paper aggregate, and leaderboard Total Scores were not retrievable. · locator: Table IV (table element 5) source
Rights
| component | license | hosting | redistribution | modification | training | Evidence and notes |
|---|---|---|---|---|---|---|
| documentation | unknown | permitted | permitted | unknown | unknown | Applies to this original summary only, not to the cited sources. |
Linked records
- cites → https://arxiv.org/html/2503.21755v1
- cites → https://arxiv.org/html/2503.21755v2
- cites → https://github.com/Vchitect/VBench/tree/fd18b3d055cb0fc6f066ca90fe2c3c8cbb698490/VBench-2.0
- cites → https://huggingface.co/spaces/Vchitect/VBench_Leaderboard
- cites → https://vchitect.github.io/VBench-2.0-project